Hello and welcome to another episode of Silly Go-Kart Design! I’m your host, Charles.
I knew that immediately after coming back from the Detroit race, I’d totally drop off the radar for two weeks as the MIT-SUTD Global Leadership Program drew to a close and everyone rushed to finish their creations. What does global leadership have to do with silly go-karts? Hell if I know, but they asked me to run it last year and it sure as hell worked out! I don’t try to read too deeply into things with “leadership” in their titles.
This year, the program was expanded to 36 students. Oh, dear.
Thirty were shipped from Singapore, while there were six MIT students who were accepted into the program after a local call for sign-ups.
Luckily, the crew for this summer was made up of some pros. First, long time compatriot in robot arms Jamison, then there was Banks (who was also an undergrad TA for the spring 2.00gokart session), and joining new this summer was Paige, who was actually a student in this year’s 2.00gokart and decided to help TA this summer session. Without highly experienced TAs, this summer would have been actually impossible.
Before I dive into the action, here are the deltas from the spring 2.00gokart. There’s some important changes that we made to scale the class to the biggest it’s ever been, now with multiple instructors. Much of the documentation was moved to The Cloud™ through Dropbox or Google Docs.
The Secret Ingredient
This past spring, I decided to spice up the design space by giving all teams the opportunity to use a
troll free part: a pair of pink Harbor Freight wheels, which has since been decided was the best idea ever. As I mentioned in the spring 2.00gokart recap, I wanted to use Vex Omniwheels as the troll free part, because why the hell not. I’d checked out Vex omniwheels in person before and they seemed to be well-constructed enough to handle vehicle loads.
I’m proud to say that the
troll design sweetener totally worked. Multiple teams went for the omniwheel solution, and none of it ended in little rollers being scattered across the parking lot.
In the three or so weeks preceding the program, I ordered a load of parts that, from past experience, students tended to gravitate towards the most.
In past classes, everyone’s mostly gone for the classic Kelly Controller and Turnigy SK3-63xx combination, a wheel or two from Monster Scooter Parts, chain drives, pedal throttles and thumb throttles alike, the weird little black brake calipers, a Hella switch, and so on. Then there’s always the odd DC motor team, so I got stuff to cater to them too, with the most common DC motor used previously (the 24V “500W” size scooter motor). This cabinet was pretty empty by the end of the class.
I try not to ‘kit the class’, but with time so crunched during the summer session, this was one way of making sure people could finish time. Teams that ordered a part I just happen to have received their parts generally after a day or two’s delay so they could get working quickly, but not right away to preserve an element of fairness. My benchmark for fast shipping is McMaster and Amazon Prime, so most teams got most of their shipped parts in the same timeframe anyway. Surplus Center folks, however, were out of luck.
It’s important to note that the contents were not made known to the teams beforehand. By the end, everyone pretty much figured out what kind of throttles and stuff I had anyway, but the work towards the beginning was reasonably independent.
Bill of Materials
One of the cornerstone elements of my ‘class’ is that all teams order their own parts (beyond the small material pile I provide them). In all past terms and semesters, this was accomplished by one team member sending me, via e-mail, a list of stuff to order. This worked well enough, especially since I am the only person with a procurement credit card.
What this system didn’t have is a way for any other instructor to verify/advise on perhaps badly informed or thought out purchases. Nor could students see ahead of time what their teammates ordered – near the end, everyone starts sharing parts and excess materials. We wanted there to be a bit more networking available to them, as well as to have the ability for each instructor to oversee the activities of a handful of teams.
For this BOM, we set up a Google Doc spreadsheet that was editable by invited students. Each team designated one member to be the go-to for finances. Through this spreadsheet, they were able to not only specify parts to be ordered, but also have a real time check of their remaining budget.
Students had made their own BOM Excel spreadsheets in past classes, but this made everything uniform and accessible.
The ordering protocol was that every purchasing day (Monday and Thursday of the 3rd through 7th weeks) I’d sweep through the BOM and then highlight green any items that were ordered. I’d highlight in red any items that were NOT ordered – the reason usually being that the specification was incomplete or the item was out of stock. Students were warned repeatedly ahead of time in both Milestone documents and the “Resources and Parts” lecture that this was going to be enforced strictly. We had very little issue with this overall.
This is an example of a fully filled out team BOM from the end of the class:
As can be seen from the image, I indicated when the order was fulfilled – for “class stock” parts, I changed the line to green when I made the “delivery”.
This BOM system worked out great, and I highly recommend it for anyone else trying this. Around the middle of the summer, one team made a “Stuff to share” tab with a similar format and quite a few teams listed their excess screws and materials.
Waterjetting and 3D Print Queueing
In semesters past, I was also the concentration point for all teams’ waterjetting files. Since 2.00gokart and last year’s GLP program only hosted 10 teams, it was still reasonable to keep track of. With competent instructors and more teams this time, we chose Dropbox as the venue for team parts submissions. I set up an instructor’s Dropbox account using our instructor email list as the account holder. The students were asked to either form a new Dropbox account or purpose an existing one (such as a personal account) for the class, and we sent invites to the e-mail addresses they gave us.
This way, any one of us could see in real-time what they uploaded. This caught a handful of potential “waterjet derps” before they were accidentally manufactured. I also recommend using this method, or an equivalent service, to sync fabrication files with instructors.
Within each team folder, students were asked to create folders with the date of submission (e.g. “Waterjet 23 July”) so we could keep track of what was submitted when. I put example folders in each group, such as “Waterjet 1 January” – this backfired at least once when one team absentmindedly put all of their waterjet files into that folder instead of one with the submission date on it, so we didn’t find it for days! Confusing examples are to be avoided in the future, but I gotta wonder some times, you guys….
So these back-end improvements are basically what differentiates this summer session from past sessions. I’m pleased with how they turned out – overall, there was minimal instructor confusion.
the build season
Now it’s time for the war stories! By all measures, this was the smoothest show we’ve pulled off yet – there were only a handful of issues that I’ll address soon. We begin with the first day:
One perk of having multiple instructors is someone can always be taking pictures. So here’s a photo of me giving the introductory lecture. In this one, details were given about the class, schedules, TA/Instructor/shop hours, design rules, and so on. Videos were shown of past semesters. That was more or less the calmest this whole summer was going to get.
Right afterwards, all hell breaks loose:
The demo parts were busted out and people began getting a sense of dread for exactly what they’ve gotten themselves into.
…and test drives were handed out. Not of Chibi-Mikuvan, mind you – that’s how people die. I put it out for display as a slight contrast against most of the aluminum extrusion framed, outrunner-powered karts present as examples (including the Chibi Twins).
It’s interesting to note that the majority of Singaporeans do not drive or own a car. The Singaporean government has purposefully made it very expensive and time-consuming to own a private vehicle on the island, due to traffic congestion and land concerns, so instead they spent massive money on a very well developed transit network. Seriously, I can get clean across the whole country in an hour on the trains, and buses will take me to the nooks and crannies. That’s how Shane and I explored so many resources when we were in Singapore. What buses don’t do, taxis will. So really, most of these students have never touched a motor vehicle in their lives.
And boy, did it show. The test drives comprised a whole lot of “100% throttle, 100% brake, 100% throttle AND brake at the same time”.
I once calculated that it would have cost about $85,000 to put mikuvan on the road in Singapore, not that they’d even let it in the country in the first place for being such a disaster. In the U.S., it cost me about $950: $800 to buy the thing, $50 for a PA temporary plate, $75 for a MA registration, and $30 for a MA state inspection sticker – ignoring the transit costs of renting a truck and trailer, of course. In singapore, the “Certificate of Entitlement” saying you may even own a car in the first place would have been around $80,000 (converted to US).
These kids will never understand the joy that is being under your vehicle, covered in burnt motor oil grunge, while contorting to European gymnastics levels to reach an impossibly tightened nut which you quickly discover you used the wrong socket driver size for. Maybe this is for the better.
In the first and second weeks, students generate ideas for their designs and perform analysis of their desired drivetrain/performance parameters to select the core components – generally system voltage, which motor, which controller, and what size wheels.
One major aspect that differed from last year was that most of these students (something like 66%) were not engineers, electrical or mechanical: they were Architecture or Design students. Many of them had never touched hardware or tools in their lives. I was not informed of this beforehand, else I would have tailored the curriculum with more introductory material. It became painfully apparently during Week 2 when the heavy drivetrain and motor math was introduced – the amount of blank stares and glazed-over looks I received was disconcerting. As a result, we had to hold more hands during “office hour” tutorial sessions. Eventually, most everyone got their heads together and understood why I was making them perform these analyses. Other “crash course” subjects that were a direct result of the students being majority not MechE (unlike last year!) were Solidworks and machine design/mechanical parts knowledge areas.
Pursuant to the lack of experience in this field, many of the karts were somewhat derivative technology-wise. Seen above are students measuring and appraising Paige’s 2.00Gokart project. Many measurements of Chibikart were also made. In the end, though, I’d say the majority of vehicles were not copy-and-paste: the students took the modular concepts of what worked from our examples and reused the modules in their own designs. For example, the Chibikart style of motor mount/corner structure was popular, but in a kart which otherwise had nothing to do with Chibikart. This is totally fine by my books.
Hey! That looks familiar!
Now we’re moving onto week 3-4. Prototyping and getting your full-size mockup together was key, here. I bought something like 300 pounds of 1/4′ and 1/8″ MDF from a local wood supplier , just like during spring 2.00gokart, to let the students prototype to their heart’s content.
Again, the difference between the ‘prototyping’ they’ve been taught and my style became apparent. During this period, some students were frustrated that they were lacking material to make the traditional “prototype” seem in college design and build courses: foamcore, cardboard, wooden dowels, and the like. A few were not sure where to start because of this.
We explained as best as possible that the idea was to go directly from detailed CAD to full-size mockup. And again, this would have been totally fine for a crew of Mechanical Engineers, but time was lacking to accommodate the desires of the Architects and Product Designers. After some frowny faces, people dealt, and the mockup frames sprouted quickly.
The MDF ran down low towards the end of the class (week 6 and on) because many teams used very generous amounts to create bodywork, custom seats, and the like; necessitating me reordering the MDF shipment. My apologies to everyone in the Cambridge-Somerville-Medford area who might have been trying to obtain 1/8″ and 1/4″ MDF materials in the past few weeks.
Hardcore fabrication begins around week 5 as peoples’ parts come in and they realize that “steel rods” don’t make “steel axles” necessarily, among other intriguing facets of mechanical engineering. Also, you don’t tap threads with Loctite as a cutting lubricant.
Above, Tinymill is seen with its groupies. We ran machine training sessions as-needed for team fabricators.
Here, a student is seen nesting in a pile of go-kart parts.
We began taking waterjet cutting submissions during week 4 (basically early July), and one Milestone requirement was to submit several different waterjet parts. Unfortunately, not having a grade or something of coercive force behind it, few teams adhered to the requirement. Submissions really began hardcore around Week 6, which resulted in plenty of waterjet spawncamping by all the instructors.
One aggravating factor was that we originally intended to split the duty between two machine shops which had waterjet machines, but one of them ended up having machine and water supply problems in the middle of the summer, causing us to have to ditch entirely to one shop. That shop was one where only I and Jamison had authority to run the machines, so we ended up having to babysit the waterjet for 2x longer than intended.
It’s important to have a reliable service supplier, and possibly even a backup supplier, if you run a class similar to this. We could not have foreseen the first shop’s issues (and frankly, neither did they), so if I did not negotiate the second shop’s permissions earlier, we’d been b0ned (that’s a technical term).
During the last weeks, the design students’ creativity really began showing through.
I actually like having this summer crew of mixed backgrounds and majors. Why? Because nobody really build a normal go-kart. Even though there were challenges, it’s refreshing to see some experimental, kind of out-there designs, instead of 4 wheels and a steering column. There were rear-steers, there were hand controls, there was an attempt at torque vectoring, and so on. The lineup looked like something out of Wacky Racers.
Nobody’s under grading pressure to “do it right”, so they do it how they want – once again, aligned with my goals for the class.
Just take a look at “Trollkart” here. This was one of the normal ones. They found out that they didn’t really have a good ergonomic location for the throttle pedal. Solution? Make it a hand throttle pedal.
Also, they used the 1 Harbor Freight pink wheel that I gave them for free as a seat cushion. As it turns out, the hub really does not project upward enough to… uhhh, cause damage. They assured me of this fact, which I later confirmed.
…and yes, we had an “Omnikart” team! It took them a ton of effort to squeeze four motors and controllers under the statutory $500 budget, but they did it . Sadly, they spent too much time getting the mechanics running and dropped from the running for the race – instead, spending the whole period during the race trying to get it to drive. Maybe this was a little too ambitious, but the team was super into it the whole time. It was simply too much work for relative novices in an 8-week build period.
I hope to see Omnikart slowly hovering around SUTD in the future.
As week 7 draws to a close, chaos breaks loose in the IDC. Lines for the cold saw are backed up out the door, and people are starting to work on vehicles wherever they can find space!
We set up the track the same way as last year and this past spring’s 2.00gokart. This procedure is now well established enough with the safety office and parking office that I think I can throw a full event with 2 weeks’ notice, which is pretty awesome. The students were all eager to finally test their vehicles – remember that for many, this was the first complete functional project they’ve worked on.
It was a hard decision for us, but at the end of the build weeks, we decided to scrap the Time/Energy contest we’ve run in years past. The reasoning was that so many teams were just out to drive around novel designs that it made little sense to try and meter their energy usage. So really it became a race of time-only, and a few teams were clearly out to just drive their creations regardless of lap time! I believe that learning is most effective when you’re not under duress, so in the spirit of the program I went along with the instructors’ opinion to scrap metering the energy usage of the teams. It removed a significant bottleneck in queueing karts up for the individual runs.
As I was busy running the event itself, these (awesome) photos are mostly courtesy of Banks and his übercamera.
A great motion anti-blur of one of the omniwheel teams. They were clearly just having fun spinning around everywhere with omniwheels in the back. I took this thing for a spin after the formal event was done, and it’s borderline impossible to control after you get up to speed!
This is what happens when you don’t listen to me about fastening your wheels on properly, guys. One of the teams had an unfortunate, chronic disease of shedding wheels.
Field repairs! I strictly enforced a no-bringing-your-own-tools rule this time, so teams had to make do with a “pit kit” we assembled of the most common hand and power tools. During 2.00gokart in the spring, there were some problems with teams sneaking tools out of the IDC because they needed a special size of long-reach ball-ended hex key to reach their wheel mounting screws or whatever. Well, this time, I had a whole lecture devoted to design-for-maintenance, so y’all have no excuses!
One of the ‘off the wall” karts: a rear-steer, tiller-controlled trike. This is basically a normal go-kart turned completely around. Their complex rear steering linkage ended up biting them, but otherwise it was highly entertaining to watch snake around the course’s cones.
One of the “normal” karts, from a team of self-proclaimed newbies. For being said newbies, they did quite well; ultimately, they shredded their two rear tires!
An issue crops up with Team Newbie Kart.
We called this one “Overconstrained-kart” because they had a single middle rear wheel with drive wheels on the sides. They insisted that the center wheel was needed to support the driver’s weight. Pretty much everyone insisted back that they’d get much better performance if they removed it, which did happen during the middle of the event.
It did perform better.
One way to be fiercely stylish is to match your custom go-kart’s seat.
Don’t worry, Omnikart did get up and running during the event, but they burnt out two of their motor controllers. If I had known that the “40A” robot controller they picked for their design were about as well-built and rated as most robot controllers on the market, I would have offered them a free pair of RageBridges.
After the individual team runs were done, we moved onto “Anarchy Hour” where everyone who wanted to run head-to-head could do so.
At this point, MITERS invaded with Chibi Atomic Jeep! Sadly, I couldn’t get Chibi-Mikuvan back up and running in time due to the necessity of running the whole show.
The Electric Vehicle Team also showed up with its motley crew of rideables. And… wait, is that LOLrioKart?! Why, yes it is – under the command of a new captain. LOLrioKart sat abandoned in MITERS for almost 3 years before it was revived this year. This is the first time it’s been on a track of any kind.
Yeah, it’s as precarious and top-heavy as I rememebered it. Even more so, with it lacking 150 pounds of nickel batteries at the very bottom!
The photo of entirety for this year. Check us out in the background acting all goofy!
the future of 2.00gokart
There were plenty of improvements on the class structure that will make future sessions that much more streamlined, including the use of Dropbox and Google Docs to coordinate instructor effort. I think the class procedures and content is very much at a plateau of stability such that the class can be redistributed easily to anyone who thinks they can run a version of it. Already, I’m seeing derivative and inspired seminars/classes pop up – from a group at UC Berkeley to the Artisan’s Asylum to individual makerspaces.
Unfortunately, I think I’m no longer in a position where I can run 2.00gokart (or the summer session) again. I’m first and foremost responsible for the daily operation of the IDC fabrication shop. It used to be that the IDC was very small and easily managed alongside teaching, but nowadays it’s grown to house dozens of researchers and students. I can no longer justify the time put off running the shop spent towards teaching the class. Each time the class runs, the shop falls into disrepair because I literally only have time to fix what’s broken just enough to keep the students working.
This summer saw my spare time stretched to nearly its limit, and at times, it was frustrating to try and keep up with researcher’s needs while answering go-kart questions. It sucks, since I want to see this happen consistently and would vountarily do so if relieved of the need to run the shop it’s housed in. I’ve let the higher-ups at SUTD know about this issue, and I’m making plans for the continuity of 2.00gokart for next spring and years to come, but for now, I see myself fully entering the role of facilitator and stepping aside on my teaching duties.
(read: IT’S YOUR DAMN PROBLEM NOW! MUAHAHAHAHAHA…)
As for the class itself, I’m once again making the materials I generated freely available online. There are two forms:
- First, a ZIP file containing only my lecture notes, “milestone” weekly report documents, and other files which I have sole copyright over: MIT-SUTD EV Design 2014 Reference Files. Not only does this have the 8-week program for the GLP, but it also contains this past spring’s 2.00gokart class Milestones, which is a 12-week program with more discrete weekly goals. This past spring, 2.007 itself implemented a “physical homework” system, which is also included in the Milestones but are not tied to or required to use the course materials.
- Second, I’ve made the MIT Stellar Course Management System page for this summer session publicly accessible (Global Leadership Program 2014). I’ll investigate maybe having this ported to OpenCourseWare for more permanence, but for now, Stellar is stable enough.
That pretty much wraps it up for this year. The smoothest and most enjoyable season of “silly go-kart design” might very well be its last, but we’ll have to see what the future holds there. The students have once again packed up their vehicles and parts for the trip home, so at the very least, we have sprinkled Singapore with even more seeds of mischeviousness!
As for myself, it’s going to be a week of cleaning up the tornado disaster that is the shop, as well as shifting my focus onto ROBOT SEASON! BECAUSE DRAGON*CON IS SOMEHOW 2 WEEKS AWAY! Stay tuned for updates on that front.